Machine learning approach for simultaneous measurement of strain and temperature using FBG sensor

In this paper, we have demonstrated a new and efficacious approach to integrating the fiber Bragg grating (FBG) by employing an Artificial Neural network (ANN) technique for measuring the temperature and strain simultaneously. The interrogation experiment has been carried out to collect the input datasets which have been used to train and test the model meticulously. We have analyzed the performance by means of different sensing parameters. The result showed that the improvement in root mean squared error (RMSE) using our proposed ANN is 680 times and 164 times for the temperature and strain measurement respectively compared to the standard transfer matrix method. In addition, a higher sensitivity is achieved using the model. Furthermore, in order to verify the proposed model, different reported sensor matrices have been taken into consideration. A remarkable improvement in the performance is attained for all cases. This new ANN-based interrogation modelling may provide a new understanding of different kinds of sensing applications with highly improved performance.